Utilizing Amazon Cloud Infrastructure for Low-Cost

advertisement
Utilizing Amazon Cloud Infrastructure for low-cost Evaluation or
Robust Production Performance of the VIVO Application
Abstract
When an institution wants to evaluate the adoption of a new
server based application, there is usually a long delay between
the initial idea and getting an environment set up for the actual
evaluation. Delays in evaluating solutions can result in a
reallocation of previously available resources or a general loss
of interest. A faster deployment solution for VIVO is needed.
Current evaluation methods require IT intervention and
approval to set up or use of a localized virtual machine (VM).
Both of these methods have their pros and cons, however with
the use of cloud services such as Amazon AWS we can
eliminate the red tape and make the evaluation available to a
wider audience than a localized VM allows. The development
of an Amazon Machine Image (AMI) that has all of the basic
services required to truly evaluate a full VIVO environment,
including VIVO, Harvester ETL toolset, Fuseki SPARQL endpoint,
and all supporting dependencies will allow us to produce a
quick and robust environment.
Introduction
During the 2010 VIVO conference the publicly available VIVO
VirtualBox image was used as teaching tool for setting up VIVO
and ingesting data. The VirtualBox image had issues, as some
users had security restrictions that did not allow them to use
the image and the image didn’t contain all of the tools that
were available for VIVO due to size constraints. The size of the
VirtualBox image, while smaller than it could be, caused
additional problems for those who hadn't downloaded it prior
to attendance at the conference. For the next conference the
Instructors needed a method to quickly create a complete VIVO
environment at low cost. With this goal in mind, the Enterprise
Software Engineering group began work on an Amazon
Machine Image (AMI) that would combine all of the most
beneficial tools and applications to provide the best evaluation
of the VIVO application. The AMI was modeled after the UF
implementation, which utilizes Amazon AWS, and allows the
institution to create an EC2 instance that fits their budgetary
and performance requirements.
Vincent J. Sposato1, Stephen V. Williams1,Narayan Raum2, Chris Barnes2, Mike Conlon3, Nicholas Rejack3,
Erik C. Schmidt3, Laura Guazzelli3, Will Colante3
1
Enterprise Software Engineering, Academic Health Center IT, University of Florida, 2 Clinical and Translational Science IT Group, Academic Health
Center IT, University of Florida, 3 Clinical and Translational Science Institute, University of Florida
Discussion
Why EC2?
The Amazon EC2 infrastructure provides a low-cost, instant
evaluation environment that is easily scalable to become your
production VIVO environment. The underlying systems that
make up the Amazon AWS ecosphere allow performance
tuning in ways that most institutions would never be able to
undertake.
On-Demand vs. Reserved
The on-demand instances on Amazon on very good for
evaluating, testing, and other cases where you are not ready to
make a significant commitment to the server infrastructure.
However, once the decision has been made to make Amazon
your production, long-term home then reserved instances
need to play a role. The reserved instances offer a significant
cost reduction per hour, and therefore a huge reduction in
overall cost of your environment. While they do require an
initial down payment, that amount does very little in the way
of affecting your overall cost savings.
Contact Us
Stephen V. Williams
Vincent Sposato
vsposato@ufl.edu drspeedo@ufl.edu
Narayan Raum
nraum@ufl.edu
Chris Barnes
cpb@ufl.edu
Mike Conlon
mconlon@ufl.edu
Nicholas Rejack
nrejack@ufl.edu
Erik Schmidt
cavedivr@ufl.edu
Laura Guazzelli
laura2@ufl.edu
Will Collante
wjc813@ufl.edu
Download